[ILUG] ATA/66 drives and IDE interface

[Paul Kelly]

Niall O Broin wrote:
> I asked a while ago about support for large IDE drives and good ATA66
> controllers. From what I received, I gather that 2.2.15 on supports big
> drives, so that's that.

That's not an ATA66 issue. ATA66 offers three things. A better cable,
66MHz data speed, and SCSI style disconnected operation & command queue
(you give a drive a list of commands to execute, tell it to go off and
do it, then you get to talk to the other drive on the same channel while
the first is busy).

> However, I got one interesting response about the
> controller, which was that in real terms the speed increase was not
> necessarily huge, and that if I wasn't in absolute need for that, I could
> use a normal IDE interface.

I know of only one drive that is substantively affected by the use of
ATA66 rather than ATA33 - the very latest IBM disks have a straight-line
speed of up to 37MBytes/sec.

> But how does ATA/66 to normal IDE work ? Do
> you need a special cable, and does the use of such a cable automagically
> tell the drive to operate in an ATA/33 mode ?

The cable is enough to let both drive and chipset know if they can use
ATA66 speeds. If you look at the 80-way ATA66 cable, you will see that
one of the wires has a small section cut out which obviously breaks the
signal on that wire. The drives and chipsets use that to recognise the
80-way cable. I believe (though i don't know for sure) that you can
still use the disconnection, but AFAIK no OS uses the feature as yet.

With a 40-way cable you can use an ATA66 drive on a regular IDE channel,
and an ATA33 drive on an ATA66 channel. It's all nice and backward
compatible that way.

Paul.